Do It Yourself


Personalize your text-to-image generation with your custom dataset

Implementation of Dreambooth↗︎ with Stable Diffusion↗︎. The approach allows a pretrained text-to-image model to be fine-tuned with a few images of a subject, such that it learns to bind a unique identifier with that subject. The unique identifier can then be used to synthesize novel images of the subject in different scenes. This technique enables the subject to be synthesized in diverse scenes, poses, views, and lighting conditions that do not appear in the reference images. The technique is applied to several tasks, including subject recontextualization, text-guided view synthesis, appearance modification, and artistic rendering.